An old adage says that if you aren’t paying for a product, you are the product. This business model is pervasive and is a good way to understand how companies, both traditional and nontraditional, make money. An easy example is social media companies (Facebook, Instagram): they provide their products for free but make money selling customer information to advertisers. For most consumers this is a net improvement over a traditional fee-for-use business, as they are willing to give up some privacy to get a good service at no cost.
We can also extend this model to companies that provide services cheaply (below cost or certainly below a market-clearing rate) in exchange for making money off of their ‘customers’ in other ways. Think of newspapers: although many charge a fee to their readers, they make additional money in advertising. Businesses following this model have varying degrees of transparency and corresponding benefit or harm for their customers. When the business is transparent, everyone benefits, but it can be inefficient or harmful when customers do not realize that they are being sold or what this costs them. One of the least transparent applications of this model, at least in finance, is the way discount brokerages make money off of their retail (non-professional) clients. Discount brokerages are online brokers (companies that allow you to place order to buy and sell shares) that compete on price and cater to unsophisticated clients with small accounts. Think of ETrade, Fidelity, or your bank’s brokerage. They typically offer low to no minimum balance and charge fixed commissions of less than ten dollars per trade. An examination of how retail trading works provides a good introduction to the current equity market structure: the environment you’re entering as a retail trader.
Payment for order flow (PFOF) is a nearly universal practice among retail brokers in which exchanges pay brokers to direct retail orders their way. This revenue stream is not huge compared to traditional fees (roughly 25% the size of trading commissions) but, as we will see, it is enough to pervert brokers’ incentives.
First, it’s necessary to understand how equity markets have evolved. The stock market has become increasingly fragmented with advent of digital exchanges and alternative trading systems (ATS). There is no single “stock market”. A broker could choose among nearly 60 venues for an order of a New York Stock Exchange listed stock. These are not only traditional exchanges, which post buy and sell orders, but also internalizers (hedge funds and wholesalers that internally match buy and sell orders), ECNs (electronic communication networks that match orders) and dark pools (exchanges where order information is hidden from traders). For simplicity, the term exchange will be used in a broad sense for all trading venues.
So, given the large number of options, how does your broker choose where to send your order? In theory, brokers should route orders to the exchange that will be the best for their clients, i.e., the venue where a trade is likely to complete quickly and at a good price. In practice, the brokers take into account the fees that they will get from each exchange when choosing between venues.
What I have just described is a classic case of the ‘principal-agent problem’. This occurs when the incentives for an agent are misaligned with the principle he is representing. In this case, brokers act as the agent for their customers, who would like them to get best execution for their orders. Brokers are instead incentivized to maximize the payout they get from exchanges. The problem is exacerbated as it is hard for traders to see choices made by their broker or understand the impact of these choices.
This raises another question: what is special about retail order flow and why do exchanges pay money for it? Retail order flow is referred to as uninformed. This means that unlike, say, a large hedge fund, retail traders are not going to move the market with their trade and are unlikely to have better-than-market information on the accurate price of a security. Essentially, in the short run, retail investors have no predictive power about the direction of price changes. To make another simplification, the majority of trades are done by market making algorithms (also known as high frequency traders or HFT) that prefer to trade against uninformed retail traders rather than informed institutional traders. Exchanges charge fees to HFT for executing against this uninformed order flow, and pay some of these fees to brokers to incentivize them to send retail trades their way.
All of this sounds very bad, but it is important to take a deeper look and to recognize that there are several reasons why it might not be as ‘bad’ as it seems.
First, despite the scary names, dark pools and HFT are not malicious entities that are ‘out to get’ mom-and-pop investors. Quite the opposite: it is generally accepted that they increase the efficiency of markets in ways that are most beneficial to uninformed traders by lowering trading costs and improving price accuracy. There is nothing inherently bad about HFT, or even the fact that they want to trade against retail order flow.
Second, we have discussed the notion of ‘best execution’ of a trade without properly defining what it means. Three factors contribute to the quality of a trade’s execution: the price that the trade is completed at, the speed of execution (how long the order takes to complete), and the fill rate (what portion of orders get completed). Brokers under the jurisdiction of the US Securities and Exchange Commission (SEC) have an obligation to provide best execution for their clients, and the SEC specifically mentions all three of the above factors . So, in theory PFOF should not enter into brokers’ calculus of where to send orders. This ideal breaks down quickly in the face of real world evidence. The requirement for best execution in terms of price is quite robust, as all exchanges are required to enforce consistent prices and must even re-route orders to other venues if a better price can be had there . It is very easy to see if brokers or exchanges violate this rule, and the SEC is quick to issue large fines for violations. However, measuring the speed of execution and the fill rate are harder for both regulators and brokers to monitor, which makes a complete enforcement of best-execution impractical. As we will see, speed of execution is a particular concern. An order that gets delayed can be executed at a worse price or, in the case of a limit order (an order with a price boundary), not at all.
The impact of weak execution speed is negligible to non-existent for the majority of trades, but a poor fill rate on limit orders is troubling. Consider a simple example. You send in a limit order to buy 100 shares of Apple with an upper limit of $100.5 each while the price is at 100$. Instead of sending your order to a venue where it would execute immediately, your brokers sends it to one with slower execution of limit orders and a poor limit-order fill rate. In the meantime the price rises through the limit and continues to increase. Your order never gets filled, and if the price rises to $110 over the next week slower execution has directly cost you $1,000 dollars.
Furthermore, in theory, if the market for exchanges was truly competitive, the fees paid by each exchange would be equal, negating the problem. Not only is this not the case, there appears to be a negative relationship between fees paid and quality of execution . Exchanges that have worse execution pay higher fees to brokers! This lends weight to a model for exchanges that is similar to the model of brokers discussed above: some exchanges compete on price (PFOF) rather than on execution quality.
This practice is nearly universal among both exchanges and brokers, but determining its impact is difficult. Even though information on the payment/fee structures of exchanges is freely available, the routing of brokers’ orders is harder to parse. One paper analyzing information from the SEC found that four of ten popular retail brokers routed all orders of a certain type to the exchange with the highest payment (and correspondingly the worst execution). Another analysis showed that the only broker to route significant volume to the exchange that appeared to offer the best execution was one that passed along a portion of the PFOF it receives to its trading customers. The size of the revenue stream gives a further indication of the significance PFOF has to brokers. For TD Ameritrade, a popular online discount brokerage, the revenue from PFOF is roughly a quarter of traditional commission revenue.
In Canada, the issue is complicated by the fact that the Canadian Securities Administration (CSA), our equivalent to the SEC, banned Canadian exchanges from paying for order flow. The CSA has since worried that this causes brokers to preferentially send the orders of Canadian traders to the US, reducing the competitiveness of Canadian exchanges and hurting traders.
Among the critics of the current system, a recommendation is widely made for the SEC to compel brokers to pass on any payments directly to their clients. If instituted this would mean that retail traders would pay higher fees to their brokers, but would get back most of the difference with rebates from exchanges. Retail traders would additionally benefit from better execution. As a result, clients would likely be made better off and brokers no worse off.
So what’s the upshot of all of this? Is the market rigged? No. Are brokers evil? No. Should mom-and-pop investors be concerned? Perhaps. Being informed is never a bad thing. Ignorance of market structure is the cause of the problem, as most brokers compete on price rather than quality of execution. As stated before, there is absolutely nothing wrong with the ‘customer as a product’ business model as long as customers know what they are giving up to get cheaper service. It might well be that customers understand all of the above and are willing to tolerate the occasional negative result of poor execution in return for paying lower commissions.
This discussion is by no means a comprehensive analysis of market structure, brokerage practices, or execution concerns and is largely simplified. However, it should be enough to introduce the idea that in trading, as with most things, cheaper is not always better.
